Economic analysis of air pollution impacts from onroad

Economic analysis of air pollution impacts from on-road mobile sources on health risks in the Jakarta Metropolitan Area A research proposal Mia Amalia 23 July 2007

www. as. wn. com www. time. com www. eia. doe. gov www. nature. com Municipal waste Soedomo et al. , 1991 www. usc. edu www. civeng. unsw. edu. au www. usc. edu

Research questions and methods 1 What is the contribution of on-road mobile sources to air pollution, represented by concentration of PM 10 and O 3, in JMA? Urban air pollution dispersion models 2 What impacts do PM 10 and O 3 have on human health? Dose-response models 3 What values do JMA citizens have for lowering health risks resulting from the decrease of air pollution concentration? Stated preference methods: Choice modelling or Contingent valuation

Urban air pollution dispersion model - definition Meteorological model: Dispersion model: § predict atmosphere’s § the simplest model is. Available models ability to disperse, dilute a Box Model to be modified: and transfer pollutants § output: concentration. SIM-AIR of § input: wind, temperature, CAMx pollutants in sub areas topography § output: transfer coefficient Q, air pollution creation rate Emission model: § energy use from every sector § output: emissions from every sector H z x y u, wind velocity W b, pollutant concentration from another box L Source: de Nevers, 2000

Urban air pollution dispersion model - strengths and weaknesses SIM AIR: Simple Interactive Model for Better Air Quality § Flexible and non-location specific. § Can model both primary and secondary PM 10 § Needs less data than other available urban air pollution models. § Needs some additional algorithm to be able to model O 3 formation. Source: Environmental Management Centre, 2006 CAMx: Comprehensive Air Quality Model with Extension § Can model both O 3 using many types of precursor substances (NOx limited or VOC limited) and both primary and secondary PM 10. § Can simulate the emission, dispersion, chemical reaction and removal of pollutan from the troposphere. § Can interpolate emissions, land use and meteorological conditions § Needs a large amount of data for inputs Source: Environ International Corporation, 2006

Urban air pollution dispersion model - application Estimate: § industrial areas § settlement areas § transportation load Digital map Sectoral data Divide areas: § into grids or § subdistricts Digital map § Emissions from every sector in each sub-area § Transfer coefficient. § Estimation of NOx proportion for O 3 and PM 10. § Estimation of secondary PM 10 from NOx and SO 2. § Estimation of O 3 concentration from NOx and VOC. § Ambient concentration for PM 10 and O 3 in sub areas. § Sector contribution for each sub area. Meteorological data to estimate transfer coefficient Source: Adopted from Environmental Management Centre, 2006 Input: amount of energy-used by four major sectors: industry, household, power-plant, transportatio Sectoral energy consumption data Estimate: § emission distribution in every sub area § emission contribution from every sector includ transportation sector Verify Ambient air monitoring data from 23 monitoring stations

Urban air pollution dispersion model - summary of data needed Data Source Digital map for the JMA Coordinating Agency for Survey And Mapping Sectoral data and emission data: Industrial emission Ministry of Environment Local Agency for Environmental Management Power plant emission Department of Geography-University of Domestic source emission Indonesia Including population distribution National Statistical Agency Number of vehicles according to the characteristics National Police Department, Local Agency for Environmental Management Meteorological data National Meteorology and Geophysics Agency Ambient air monitoring data Ministry of Environment

Dose-response model - properties Definition: Dose-response model is a mathematical model to estimate the amount of pollutant do number of sicknesses or deaths related to a particular pollutant. Source: Kunzli et al. , 2000; Mc. Cubbin and Delucci, 1996; Hall et al. , 19 Functional forms: § log linear § logistic § Poisson regression Source: Kunzli et al. , 2000 Examples: § minor restricted activity days caused by O 3 § restricted activity day caused by PM 10 Source: Hall et al. , 1992 Strengths and weaknesses: § Can estimate number of health incidences related to a pollutant with respect to the s specific condition. § More reliable than using available dose-response function adopted from other resea conducted in other sites. § The process treated all information as uniform, cannot differentiate data based on th of health incidence.

Dose-response model - application Identification of: Regression analysis: § Health problems associated with§ Possible functions: PM 10 and O 3 asthma=f([O 3], sosioeconomic group, age group § Socioeconomic groups asthma=f([PM 10], socioeconomic group, age gro § Age groups premature death=f([PM 10], socioeconomic grou age group) § Annual incidence of respiratory § Regression of pollutant with health impacts to cer related diseases group of population to develop dose response mo § Annual incidence of premature deaths § Apply all possible functions: linear-linear, logistic or Poisson regres § Select the most suitable function based on statisti Dose-response models: evidence. for asthma and premature death caused by O 3 and PM 10 Annual PM 10 and O 3 concentration from all sub areas in the JMA – results from the first research question Source: Adopted from Kessel, 2006

Dose-response model - summary of data needed Data Source Annual incidence of Susenas data from National Statistical Agency, respiratory related Indonesia Family Life Survey – Rand Corporation diseases especially Ministry of Health, asthma National Institute of Health Research and Development – Annual premature MOH mortality Local health agency, Public hospital specialising in respiratory related diseases Related researches Annual PM 10 and O 3 concentration in every sub areas Results from urban air pollution dispersion model: PM 10 and O 3 concentration in every sub-area

Choice modelling - properties Definition: § a technique where the good in question is described in terms of its attributes and levels of the attributes. § ‘Provide a wealth of information on the willingness of respondents to make trade offs the individual attributes’ Strengths and weaknesses: § Can measure use, passive and non use values § Can evaluate several changes and focus on trade offs between attributes. § Reliable to estimate marginal value of each attribute § Has the ability to control unobservable consumer utility and lead to a better understa respondent choices § WTP is indirectly estimated from the questionnaire not by directly asking the respon § Can reduce framing problems § Still suffers from scoping problems and hypothetical bias § Complex and multiple choices can lead to respondents’ fatigue leading to irrational c § Discrepancies between the ‘whole’ value of good with the sum of the ‘part’ values § CM estimations are usually higher than CV estimations Source: Boxall et al. , 1996; Wang et al. , 2006; Hanley et al. , 2001; Blamey et al. , 1999; Riera, 2001; Bennett et al. , 2004; Rolfe and Bennett, 2000; Rolfe et al. , 2000; Mogas et al. , 2006; Bennet and Blamey, 2001

Choice modelling - application Questionnaire development Survey Data analysis using possible models: § Multinomial logit model § Multinomial probit model § Nested logit model § Random parameter logit Output: The JMA citizens’ WTP for lower health risks Source: Adopted from Blamey et al. , 1999; Hanley et al. , 2001

Choice modelling - questionnaire development Background for focus group discussion: Focus group discussions: § Link possible policy scenario for the § Elaborate background information. transportation sector to reduce health risks § Develop possible scenarios. § The status quo alternative is current condition § Choose possible attributes. without new policy for the transportation sector. § Expose possible attributes’ levels. Using results from 1 st and 2 nd questions Possible attributes and attributes’ levels: § citizens’ health conditions – results from 2 nd q § possible amount of payment Questionnaire test: § Focus group § In research site by the enumerators Source: Questionnaire: § Introduction § Framing § Statement of the issue § Choice sets § Socioeconomic questions Adopted from Morrison and Bennett, 2004; Blamey et al. , 1999; Bennett and Blamey, 2001; Hanley et al. , 2001; Wang et al. , 2006; M Bennett et al. , 2004; Boxall et al. 1996

Choice modelling - survey Sampling design: Classification: § based on districts/municipalities: 23 -97/subareas § based on socioeconomics groups: 200/groups Households are identified based on the National Socio-Economic Household Survey 2005 (Susenas 2005) Population for Susenas 2005 12 municipalities No. of blocks surveyed: and districts 624 CM Survey 23, 603, 977 Source: Supas 2005 600 Survey technique: Face-to-face interviews Possible number of enumerators: 10 Number of days needed: 15 working days (can include the weekend) District/Municipality No of Sample s Jakarta Selatan (M) 51 Jakarta Timur (M) 61 Jakarta Pusat (M) 23 Jakarta Barat (M) 53 Jakarta Utara (M) 37 Bogor (D) 97 Bogor (M) 23 Bekasi (D) 50 Bekasi (M) 51 Depok (M) 35 Tangerang (D) 83 Tangerang (M) 37 Source: Adopted from Robson, 2004; Tacconi, 2006; Bennett and Adamowics, 2001; Gordon et al. , 2001; Keller, 2005; Wang et al. , 2006; Mitchel and Carson

Contingent valuation - properties Definition: § a technique of obtaining values by using a survey method. § directly ask people’s WTP of a good in question. Strengths and weaknesses: § can estimate all types of environmental values, including non-use values § reliable for collecting information on the individual WTP for public infrastructure proje and public services in developing countries § can be used among a poor and illiterate population and can obtain a consistent answ § the application is limited to up to two policy alternatives § suffers from biases such as strategic, starting point, hypothetical and interviewer bia leading to WTP estimate bias. Source: Tietenberg, 2006; Mogas et al. , 2006; Hanley et al. , 2001; Rolfe et al. , 2000; Mitchell and Carson, 1989; Blamey et al. , 1999; Whittington et al. , 1990; Riera, 2001; Garrod and Willis, 1999; Boardman et al. , 2006; Lechner et al. , 2003; Cameron and Quiggin, 1994; Poe, 2006; Hanley and

Contingent valuation - application Questionnaire design Possible background information: § background information, description of new government § hypothetical market, program to improve JMA’s § payment vehicle, air quality § WTP questions, § protest identification and The good in question: Health condition – answer to 2 nd § socio economics question Questionnaire is designed through a focus group discussion. Possible payment vehicle: Results from 1 st and 2 nd research property or income tax or other questions are used form of payment suggested by for focus group discussion and focus group discussion background information Questionnaire pre test § In the 2 nd focus group § In the field by the interviewer Survey procedure Same with CM Output: The JMA citizens’ WTP for lower health risks Data analysis using possible models § Logit § Probit Source: Tietenberg, 2006; Mogas et al. , 2006; Hanley et al. , 2001; Rolfe et al. , 2000; Mitchell and Carson, 1989; Blamey et al. , 1999; Whittington et al. , 1990; Riera, 2001; Garrod and Willis, 1999; Boardman et al. , 2006; Lechner et al. , 2003; Cameron and Quiggin, 1994; Poe, 2006; Hanley and Spash, 1993; Satterfield and Kalof, 2005; Whittington, 1996; Whittington, 2002

Field work schedule Tasks Secondary data collection Primary data collection Tasks after secondary data collection: Urban dispersion model building Dose response model building Specific tasks for primary data collection: Focus group discussions Sample construction Questionnaire development Enumerators recruitment First test in second focus group Second test by the enumerators Au g Se p Oc No t v De c Ja n Fe b Ma Ap Ma r r y

Plan for thesis Chapter 8 Reveal the results to assist with answering Question One, establishing the relationship between the transportation sector and ambient air pollution, in particular PM 10 and O 3 concentrations. Chapter 9 Reveal the results to assist with answering Question Two, linking air pollution with health impacts using dose-response models. ‘Dose(s)’ are based on output from Chapter 8. Chapter 10 Reveal the results to assist with answering Question Three, using the choice modelling method to estimate the JMA citizens’ WTP for cleaner air. Attributes are identified using output from Chapter 8 and 9. Chapter 11 Analyse the relevance of the research outputs with possible air pollution control policies. The benefits estimated to control pollutants will be compared with the cost of policy implementation proposed to achieved improved air quality in the JMA. Chapter 12 Conclusion